This is a pytorch implementation of the model Domain Separation Networks
- Pytorch 0.4.0
- Python 2.7
python train.py
Note that this model is very sensitive to the loss weight, our implementation cannot perform as perfect as the original paper, so be careful when you tune parameters for other datasets. Moreover, this model may not be suitable for real nature image, cause the private and shared feature of nature image are more complicated, so that difference loss cannot adapt well
We only conduct the experiments from mnist to mnist_m, the target accuracy of our implementation is about 77% (original paper ~83%), and some results are shown as follows, from left to right: recovery from shared+private, shared and private features.